Good day, guys.

I've set up web replication of my database, some articles exceeds 4GB in size.

So, web replication never ends, I ran ProcMonitor on my IIS machine, and found that w3wp cannot read my first large article, operation "Read File" gets to the offset 4294901760 and starts over from 0. Seems as unsigned int32 overflow.

replisapi.dll version is 2011.110.3000.0

Have it been fixed in further releases?

Hi Mizax,

Based on my research, changes are pulled from the database and bundled as an XML file in Web synchronization, and the maximum size for the XML file is still 4 GB in the latest SQL Server 2016 version. But replication synchronizes the changes from that file in batches. The maximum batch size of data and metadata is 25 MB. You must ensure that the data in each batch does not exceed approximately 20 MB, which allows for metadata and any other overhead.

Batch size for merge replication is measured in generations, which are collections of changes per article. The number of generations in a batch is specified by using theDownloadGenerationsPerBatch and UploadGenerationsPerBatch parameters of the Merge Agent. For large volumes of data, please specify a small number for each of the batching parameters.

If you concern about the XML file limitation, you can submit a feedback to the Microsoft Connect at this link https://connect.microsoft.com/SQLServer/Feedback . Your feedback enables Microsoft to offer the best software and deliver superior services.

Reference:
Configure Web Synchronization

Thanks,
Lydia Zhang